This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]


On Tue, May 15, 2001 at 06:02:53PM -0700, Mark Mitchell wrote:
> This patch fixes eb91.C on Solaris, a regression from GCC 2.95.x.  We
> really need to use CPLUSCPLUS_CPP_SPEC, not os_defines.h, to define
> _LARGEFILE_SOURCE, etc., since we don't control inclusion order.

os_defines.h predates CPLUSCPLUS_CPP_SPEC by many moons; I don't think
anyone would have a problem with moving from one to the other en masse.

Somebody once mentioned the possibility of having cc1plus always initially
include os_defines.h, period, for every compilation.  Thus _LARGEFILE_SOURCE
and whatnot would always be defined.  This would allow easier maintainance,
for one thing, compared to messing with specs.

A friend of mine implemented this, sortof, as a beginner's project.
It seems simple enough to do, but was breaking local builds, so we never
cleaned it up for submission.

Is this a viable idea?

> (Actually, in the long term, we need to make V3 tolerant of whatever
> combination of macros the user wants to use, but that's another
> problem.)

Yes and yes.


pedwards at disaster dot jaj dot com  |  pme at sources dot redhat dot com
devphil at several other less interesting addresses in various dot domains
The gods do not protect fools.  Fools are protected by more capable fools.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]